[Coolaler] AMD Piledriver FX Engineering Sample Benchmarks

[Olikan]

From these benchmarks it seems that AMD lost all the design and technical documents on Piledriver after releasing Trinity and resorted to putting their engineering team in a cage with an electron microscope, a Trinity chip, lifetime supply of Jaegermeister and a parrot. In the end, they produced something similar to a Piledriver module if you squint a bit and the parrot is now the lead engineer due to winning a drunken engineering swearoff.

That or this ES just might not be the final one

no...it's the future jaguar core, at 9W TDP

 

[Phynaz]

Looks like a nice performance boost good job AMD

4% - 5% performance boost = unnoticeable.

Shall we look up what you had had to say about Ivy Bridge? Oh we know your story, Intel gets 5% improvement they have "flat-lined", AMD gets a 5% performance increase it's a good job.

 

[Arzachel]

When was the last time that ES samples performed differently from retail? Never?

Trinity and Vishera uses the same(or slightly tweaked) modules. The former performs better than the later, setting all kinds of red flags about the validity of the benches.

 

[ShintaiDK]

Trinity and Vishera uses the same(or slightly tweaked) modules. The former performs better than the later, setting all kinds of red flags about the validity of the benches.

1M/2T Trinity can also outperform 2M/4T Trinity at same clocks. Go figure.

 

[nehalem256]

Trinity and Vishera uses the same(or slightly tweaked) modules. The former performs better than the later, setting all kinds of red flags about the validity of the benches.

Maybe adding the L3 cache hurt performance? Obviously AMD just adds it on for decoration

Or maybe more likely the benches are invalid.

 

[Sweepr]

I expected to see actual analysis and rational discussion on that topic, but instead all I saw was even more disturbed AMD fanboys than I see here.

Honestly, this post just kind of scares me. This guy is literally crying because AMD's new processor will be 5% better than its old processor, which still means it fails to compete with AMD's 1090t from April of 2010.

This one is kinda scary too:

Nehalem IPC + overclockable to at least 5 GHz combined with (hopefully) low price = win

 

[exar333]

Coolaler has been pretty reliable in the past, I would be surprised if these results were fabricated.

 

[Arzachel]

Coolaler has been pretty reliable in the past, I would be surprised if these results were fabricated.

I don't doubt that the results are real. Rather people are misatributing them.

 

[looncraz]

This is obviously an ES for a new bulldozer stepping. OR-C0 wouldn't be a piledriver stepping, now would it? Nope, try the FX-8150p or a relative - which I don't think was released due to some issues... if it had ever been intended for release.

This is likely nothing more than a Bulldozer with the clock mesh installed - and possibly some issues related to that.

 

[jhu]

Ugh. The Google translated version isn't very good because the posts from the original website are using the wrong words for certain terms.

 

[Olikan]

I don't doubt that the results are real. Rather people are misatributing them.

...dude this results make Vishera worser than Bulldozer, the only better result is cinebench

lol, that would make piledriver in the hall of shame...be the first time that an arquitecture tweak generate a worse chip than the original

 

[jhu]

...dude this results make Vishera worser than Bulldozer, the only better result is cinebench

lol, that would make piledriver in the hall of shame...be the first time that an arquitecture tweak generate a worse chip than the original

Actually that would be the second time: Pentium 4 Northwood to Prescott.

 

[Rvenger]

If PD was worse than BD, then that would probably be the end for AMD in the CPU market.

 

[Jacky60]

Why release products that are barely any faster than previous ones? AMD should pull out of the CPU business if this is all they can offer it reminds me of 3DFX Voodoo 4/5 hopeless showing and also when AMD beat Intel to 1Ghz then sat on their hands for two years trying and failing to twiddle their thumbs! When they released 'new' cpu's 50Mhz faster than previous model (something Intel are nearly doing now) they were again taking the p**s. They need to hire some talent or leave the party.

 

[nehalem256]

Why release products that are barely any faster than previous ones? AMD should pull out of the CPU business if this is all they can offer it reminds me of 3DFX Voodoo 4/5 hopeless showing and also when AMD beat Intel to 1Ghz then sat on their hands for two years trying and failing to twiddle their thumbs! When they released 'new' cpu's 50Mhz faster than previous model (something Intel are nearly doing now) they were again taking the p**s. They need to hire some talent or leave the party.

What. Between March 2000 when the Athlon 1GHz was released and march 2002

they

1.) Integrated L2 cache (Thunderbird core)
2.) Increased bus speed to 266 and added support for DDR
3.) released an improved Athlon (Palomino core)
4.) Oh and increased clockspeed to 1733Mhz

 

[Hypertag]

Why release products that are barely any faster than previous ones? AMD should pull out of the CPU business if this is all they can offer it reminds me of 3DFX Voodoo 4/5 hopeless showing and also when AMD beat Intel to 1Ghz then sat on their hands for two years trying and failing to twiddle their thumbs! When they released 'new' cpu's 50Mhz faster than previous model (something Intel are nearly doing now) they were again taking the p**s. They need to hire some talent or leave the party.

Why would the leave the market? Look at all of those AMD fanboys willing to pay $250-500 for an AMD processor that can't compete with the 1090t.

I mean honestly, why not charge $25,000 per processor?

 

[inf64]

Why would the leave the market? Look at all of those AMD fanboys willing to pay $250-500 for an AMD processor that can't compete with the 1090t.

I mean honestly, why not charge $25,000 per processor?

Which processor can't compete with 1090T? FX8150 is overall faster chip than 1100T,so I don't know what fantasy reviews you get your data from.

PS Don't please use "IPC" to prove your point since it alone is meaningless(and not uniform metric since it varies a lot across a range of desktop workloads).

 

[exar333]

Which processor can't compete with 1090T? FX8150 is overall faster chip than 1100T,so I don't know what fantasy reviews you get your data from.

PS Don't please use "IPC" to prove your point since it alone is meaningless(and not uniform metric since it varies a lot across a range of desktop workloads).

Stock maybe. I would rather own an x6 to OC it to get better overall performance over a BD option.

 

[SlowSpyder]

Why would the leave the market? Look at all of those AMD fanboys willing to pay $250-500 for an AMD processor that can't compete with the 1090t.

I mean honestly, why not charge $25,000 per processor?

AMD's highest priced consumer chip is $189.99 on Newegg. Your numbers seem a bit off.

 

[Ferzerp]

He's saying that there are people who would pay $250-$500 regardless of performance.

 

[BallaTheFeared]

No but really, Anand should just put a clamp on these threads before it turns into another Bulldozer fiasco where IPC increases and butthurt ensues.

 

[Black96ws6]

No but really, Anand should just put a clamp on these threads before it turns into another Bulldozer fiasco where IPC increases and butthurt ensues.

According to these graphs IPC seems WORSE lol:
http://www.overclock.net/t/1291114/...gineering-sample-benchmarks/240#post_17885199

That can't be right though!

 

[Vesku]

Which seems very off base considering AMD had to offer rebates on the 8120 and 8150 pretty soon after launch. In fact AMD being considered the value x86 producer puts more price pressure on their products, at least when not clearly leading in performance, imo.

Personally, this 1090T in my main machine has been holding up well. If Piledriver turns out to be mainly a few 100MHz more in clockspeed and not much else, then I'll just wait and see how Haswell matches the hype.

He's saying that there are people who would pay $250-$500 regardless of performance.

 

[N4g4rok]

Regardless of what potential the architecture may have had, it may be time to cut their losses and move onto a new architecture, or out of mainstream CPU market entirely. otherwise, they'll kind of limp off and dissipate when they run out of R&D money. Not sure who would buy AMD if they hit rock bottom.

I don't suppose there's a chance they'll start manufacturing some X6's again is there? it'd be nice to pick up a cheap one for a minor upgrade.

 

[Gs dewd]

So does this mean that people, like me that are still running x6's are going to hold onto there systems a little longer? Glad I have 2 x6's and 3 x4's systems to play with until something else comes out.